Near-Automated Estimate of City Nitrogen Oxides Emissions Applied to South and Southeast Asia

Cities in South and Southeast Asia are developing rapidly without routine, up-to-date knowledge of air pollutant precursor emissions. This data deficit can potentially be addressed for nitrogen oxides (NO_x) by deriving city NO_x emissions from satellite observations of nitrogen dioxide (NO₂) sampled under windy conditions. NO₂ plumes of isolated cities are aligned along a consistent wind-rotated direction and a best-fit Gaussian is applied to estimate emissions. This approach currently relies on non-standardized choice of upwind, downwind, and across-wind distances from the city center, resulting in fits that often fail or yield non-physical parameters. Here, we propose an automated approach that defines many combinations of distances yielding 54 distinct sampling boxes that we test with TROPOspheric Monitoring Instrument (TROPOMI) NO₂ observations over 19 isolated cities in South and Southeast Asia. Our approach is efficient, uses open-source software, is adaptable to many cities, standardizes and eliminates sensitivity to sampling box choice, increases success of deriving emissions from 40% to 60% with one sampling box to 100% (all 19 cities) with 54, and yields emissions consistent with the current manual approach. We estimate that the annual emissions range from 15 ± 5 mol s⁻¹ for Bangalore (India) to 125 ± 41 mol s⁻¹ for Dhaka (Bangladesh). With enhanced success of deriving top-down emissions, we find support from comparison to past studies and inventory estimates that top-down emissions may be biased, as the method does not adequately account for spatial and seasonal variability in NO_x photochemistry. Further methodological development is needed for enhanced accuracy and use to derive sub-annual emissions.